COMPUTER OPERATIONS OF THE SEQUENCE ANALYSIS OF 13 TRACE GASES IN AIR.
In order to gain some insight into how RIAS works, let us look at details of what the computer sees and does during the analysis of an air sample. Much of this detail can be seen on the computer if one carries out the measurements using the action button Analyze SC. We show here, plotted out on paper, spectra that the computer works with, with indications of the zeroing regions and integration regions.
We consider the spectrum of a sample of air captured in the District of Columbia on a weekday morning. We measure the sample for thirteen trace gases. The measurements and subtractions that the computer would do during an automatic sequence are illustrated for individual compounds in Figures 1 through 16. The reference spectra that are shown in the figures are marked with a double-lined box to show the integration area and with a heavy black dot or oval to show where the zero level was obtained.
FIGURE 1. In Figure 1. we illustrate the removal of water from the air spectrum. The reference spectrum was made on humidified nitrogen. The figure shows the zero region and integration region by oval and box. The absorption lines in the bottom spectrum are almost all due to the residual carbon dioxide. If the water removal were part of an automatic sequence, only the starting and ending spectra would be shown; the details given here would be hidden.
FIGURE 2. Figure 2 shows operations on the water-free spectrum in a region where CO2 absorption dominates. Note the wide zeroing and integrating regions. The CO2 amount of 404 PPM would be printed out at the end of an automatic sequence. There is one residual band at 729 wavenumbers. This band is due to acetylene.
FIGURE 3. The acetylene measurement and removal are shown in Figure 3. The choice of a wide zeroing region behind the acetylene band averages the residual pattern left by the subtraction of CO2. The amount of acetylene is only 0.026 PPM, giving a CO2-to-acetylene ratio of about 16,000-to-1. The RIAS technique is able to give the correct answers for both compounds regardless of the order in which they are measured. If you measure CO2 first, you get the correct answers the first time through. If you measure acetylene first, the CO2 interference causes you to get a number that is far from correct. After the CO2 is measured and subtracted, however, a second run corrects the acetylene error. The sum of the two runs then gives the correct total. One can verify this by creating and running 2-compound sequences in both the forward and reverse order.
FIGURE 4. In Figure 4. we go to the C-H stretch bands at the other end of the air spectrum. The water subtraction reveals the methane, and the methane subtraction reveals a weak residual due to non-methane hydrocarbon. Note that the methane measurement is made on a single methane feature that is separate from the water. This means that we could have measured the methane without subtracting the water lines. Note also that the zeroing was done at the top of the methane feature. If there were a large amount of non-methane hydrocarbon in the sample, this methane feature would have appeared on a slope, and taking the zero on the top of the line would have minimized error due to that slope. Error due to slope is also eliminated in a second run after the non-methane hydrocarbon is measured and subtracted.
FIGURE 5. In Figure 5 we see an enlargement of the residual absorbance after water and methane are subtracted. We measure this as Exxon gasoline vapor. We see that the difference spectrum is now quite flat in the C-H region. Apparently, the non-methane hydrocarbon in the air was mostly gasoline vapor.
FIGURE 6. Figure 6 shows operations in the spectral region where we choose to measure CO and N2O. The top spectrum shows absorption by H2O, CO2, N2O and CO. The second spectrum has the water lines subtracted away. The third spectrum has both water and CO2 subtracted away.
FIGURE 7. The air spectrum with water and CO2 removed is shown again at the top of Figure 7, which illustrates CO measurement and removal. Some artifacts remain after CO removal because of a slight mis-match of line shapes between sample and reference spectra, but these ups and downs balance each other out and do not indicate a serious error in the measurement.
FIGURE 8. After CO subtraction, the N2O absorption remains, as shown at the top of Figure 8. N2O is measured and subtracted away by means of the reference spectrum shown, using the integration and zeroing regions indicated. Since the N2O reference spectrum has slightly better resolution than the N2O band in the sample spectrum, the subtraction has left an up-and-down pattern in the difference spectrum. As in the case of CO however, the pattern is not related to an error in the measurement. When averaged over the spectral region shown, the difference spectrum is quite flat. The measured N2O concentration of 0.32 PPM is within a few percent of the known background concentration of nitrous oxide in the atmosphere.
FIGURE 9. This figure shows how the nitric oxide lines are mixed with much stronger water features. It also shows that between the water lines near 1900 cm-1 there is a usable nitric oxide line. The air spectra with and without water are both plotted with the indicated absorbance scale, but they are separated for clarity.
FIGURE 10. Here we see expanded plots of the water-free air spectrum in the nitric oxide region. The integration area and the zeroing region are marked. In the difference spectrum one can verify the removal of other nitric oxide lines in addition to the line at 1900. The absorbance scale shown applies to all three spectra, separated for clarity.
FIGURE 11. The upper spectrum of Figure 11 shows a part of the open space in the middle of the major band of atmospheric water vapor. For clarity, this spectrum is slightly displaced upwards from the residual spectrum after water is removed. The NO2 spectrum at the bottom shows that the most favorable features for measurement are between 1597 and 1600 cm-1. A favorable region for taking the zero level is between 1584 and 1588.
FIGURE 12. In Figure 12 the nitrogen dioxide measurement and removal are illustrated. All three spectra in the figure have the same absorbance scale, but are separated for clarity. Because the NO2 spectrum is quite strong, we have obtained a reliable measurement even though the NO2 amount was only 0.0026 PPM.
FIGURE 13. Figure 13 illustrates the measurement of a compound--dichlorodifluoromethane--that was present in the air sample at only one part-per-billion (0.001 PPM). Note the use of the top of the line for zeroing.
FIGURE 14. The methanol measurement in Figure 14 illustrates the ability of the RIAS technique to measure a compound even when its spectral feature appears on top of an absorption band of some unknown compound. The band in the difference spectrum is due to some compound not yet identified. This unknown compound might have been present in the ambient air, or it might have been an emanation from the plastic bag used to gather and transport the sample.
FIGURE 15. This figure shows the measurement of ethylene, with the correct concentration being obtained even though there is an underlying band due to an unidentified compound.
FIGURE 16. Figure 16 illustrates the measurement of propylene and isobutylene in the air, even though the concentration of each was only a few parts-per-billion.